Anthropic’s Haiku 4.5 Outpaces Sonnet 4 While Google’s C2S Scale Revolutionizes Cancer Research
AI Model Releases & Updates
OpenAI Expands Sora’s Video Generation to 15-25 Seconds:
OpenAI has updated Sora, extending video generation limits to 15 seconds for regular users and 25 seconds for Pro users, enhancing creative flexibility in AI-driven content creation.
OpenAI’s New Multimodal Model "gpt-5-image" Spotted on OpenRouter:
A previously unannounced OpenAI model, "gpt-5-image", was discovered on OpenRouter, hinting at advancements in image-related or multimodal AI capabilities.
Anthropic Launches Haiku 4.5: Faster, Cheaper, and More Powerful Than Sonnet 4:
Anthropic’s Haiku 4.5 outperforms Sonnet 4 in performance while offering lower costs and significantly higher tokens-per-second throughput, ideal for coding and high-efficiency tasks.
AI in Healthcare & Scientific Research
Google & Yale Release C2S Scale: A Gemma-Based Model for Cancer Cell Analysis:
The 27B-parameter C2S Scale model, built on Gemma 2, generates experimentally validated hypotheses about cancer cell behavior, advancing AI-driven medical research. The model is open-source and available on Hugging Face.
Personalized & Niche AI Applications
Matthew McConaughey-Inspired LLaMA Agent for Document-Specific Responses:
A fine-tuned Llama model was developed to act as a private AI agent, ingesting and referencing user-uploaded documents exclusively—demonstrating potential for customized, knowledge-restricted AI assistants.